Science1 distinct publisher2 min readPublished
Andreas Bauder has drilled mass-balance stakes into Swiss glaciers for twenty years. This summer the melt released them, and the monitoring network had to be repaired while it was being read.
The Scientist · Science desk

Compiled by The ScientistSomething wrong?How this is made
An ablation stake is a bet on the coming summer: drill an aluminium tube deep enough into the ice that it is still standing in September, then read off how much of it the melt has exposed. Bauder has been placing them for twenty years, and his own account of the method is that he drills them deep enough to survive, except under conditions like this summer's [1][3]. The tube he pulled out of the Rhone was 4 metres long and the glacier's surface dropped by more than 9 [2][4], so the bet lost by better than a factor of two [1].
Re-drilling in the field fixes the immediate problem. It does not fix the siting. On a steep icefall at almost 2600 metres, where summer melt rarely reached in previous decades, the team found a second stake going over and put a new one in [5][6]. The melt line this year is above 3100 metres and climbing, which converts at least 500 metres of former accumulation zone into ablation zone [6][2]. That upper ground is what the whole accounting rests on: snow compacting into ice at the top, flowing down to offset the loss below [19]. Stakes were not up there because nothing much was leaving.
The season also began about twelve weeks early. Glacier loss day, the date summer melt overtakes winter snowfall in the Swiss Alps, fell on 2 July against a late-September norm [7][3], after a winter with scarce snow and a May heatwave [8]. Some of the loss is being driven from underneath. Ilaria Santin of ETH Zurich attributes the hole torn in the glacier's lower flank to meltwater draining down vertical shafts and building a subglacial river that collapsed the ice above it, and says more water at the bed means more ice going quickly [16][17].
The 6 per cent is the number worth keeping [10]. Straight-line arithmetic on that rate, which is a bound and not a projection, empties the Alps of ice in about 17 years [5]. On the same crude basis, Switzerland held roughly 2500 glaciers half a century ago, has lost more than 1000 of them, and the 30 the Swiss media expect to go this summer are around 2 per cent of the survivors [13][4]. The Rhone itself has pulled almost 2 kilometres back up its valley since it appeared in Goldfinger, far enough that the hotel built to look at it stands abandoned [18]. Anything sized to the melt rates of the last twenty years, a 4-metre tube included, is now sized wrong.
Ranked by verification strength, evidence, and original report placement.
Andreas Bauder at ETH Zurich in Switzerland has spent two decades drilling stakes into glaciers to measure how much of the surface melts away during the warm months.
On the Rhone glacier, a 4-metre aluminium tube had emerged so far from the ice that it had nearly tipped over, and Bauder pulled it out.
Bauder: "I make sure that they're drilled sufficiently that they survive. Except when we have some special conditions like this summer... There was too much melt."
More than 9 metres of ice have melted off the surface of the Rhone glacier this season.
At almost 2600 metres, up a steep icefall, the team found another melt stake falling over; Bauder pulled it out and drilled in a new one.
In previous decades the summer melt rarely reached the point at almost 2600 metres; this year it has crept above 3100 metres and counting.
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Direct field measurement, one institution
The core observations are physical and specific - stake length versus surface lowering, melt-line elevation, glacier loss day date, tonnage lost - and are attributed to named researchers at ETH Zurich and the University of Innsbruck within an established >100-glacier monitoring programme. What holds the score below the high band is that all of it reaches us through a single publisher and a single field visit, with no uncertainty ranges, no closed-season mass-balance record, and no independent group corroborating the numbers.
Instrumented network in use; visible operational responses
Adoption here is the real-world uptake of and reaction to the melt signal. There is concrete evidence of an operating measurement deployment being read and repaired in-season across a multi-glacier network, and of third parties acting - a year-round ski area suspended and the Chamonix ice tunnel closed. Broader uptake, such as the reservoir-building adaptation the article raises, is described only as something Europe 'will have to consider', with no committed programmes, so the measured level stays mid-range.
Slightly overstated framing, hedges kept in text
The headline framing of 'on track for record-breaking ice loss' runs marginally ahead of what the season can yet support: the record outcome is explicitly conditional on another heatwave and Prinz says there is 'no final answer'. Against that, the observed quantities - stakes released, >9 m of lowering, 2 July loss day, 30 Mt lost, the collapsed flank - are as strong as the framing implies, and the article preserves its own hedges and the 2100 policy projection. The overstatement is therefore small, and the one genuinely inflated reading in the ledger, straight-line exhaustion of Alpine ice in about 17 years, is arithmetic the source does not assert.
Moderate: outlet colour and institutional stake, data-led core
The publisher has a clear attention incentive and uses it - the Goldfinger set-piece, the 'massacre' framing, Olympic-pool conversions - and the quoted researchers have an institutional interest in the visibility and funding of long-term glacier monitoring. Those pressures are visible but bounded: the quantities are specific and checkable, the experts' own statements are cautious rather than promotional, and no commercial product, vendor or funding round is being sold in the piece.
Well-grounded but uncorroborated and mid-season
Confidence is supported by the physical, first-hand nature of the reporting and four named researchers across two institutions, and limited by three things: a single publisher with no cross-outlet check, a source body that is truncated before its close, and a season that was still open at publication, so the record claim and the 6 per cent figure remain provisional.
product
Canva's forecast cut turns model routing into a product line item1 distinct publisher
science
Sleep tech turns rest into evidence, and employers are the ones who will need rules1 distinct publisher
science
Junk DNA with a day job: fly chromosomes pair by matching repeat barcodes1 distinct publisher
invest
SoftBank's $200M Buys A Retrofit Kit, Not A New Excavator1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 26, 2026